Climate monitoring is a fundamental parameter for various fields of knowledge, as
climate changes directly impact peoples quality of life. In this study, low cost sensors for
temperature, relative air humidity, atmospheric air pressure, air temperature, and altitude were
utilized. These measurements were compared with commercial sensors for temperature and
relative air humidity (HOBO®, pendant); atmospheric air pressure (HOBO®, RX3000); and
air temperature (HOBO®, RX3000). The low cost ESP32-S weather station was installed on
the campus of the Instituto Nacional de Pesquisas da Amazônia (INPA) to compare its data
with the 80m station data from the IETE monitoring network, verifying the measurements and
attesting to their accuracy. To assess the performance of ESP32-S sensors, descriptive
statistics (MBE), Root Mean Square Error (RMSE), and the coefficient of determination (R²)
were applied. From the analysis of the obtained data, it was observed that the temperature
values from the low cost station underestimated both daily maximum and minimum values,
with an MBE of -0.59 °C and a spread of 0.59°C (RMSE). Humidity showed differences of
21.03%, with a spread of 19.47% in both the temporal series and daily cycle. Pressure values
consistently underestimated throughout the daily cycle, where atmospheric pressure measured
by BMP280 captured the pressure variability but overestimated compared to the measurement
by HWS S-BPA-CM10, with an MBE of 2.8 hPa and a spread of 2.8 hPa. Although the
observed error for humidity has a larger amplitude of over 20%, the agreement between the
data is still strong with a value of 0.9. The agreement for air temperature and atmospheric pressure showed values of 1 and 0.99, respectively. Therefore, the use of a low cost weather station can be considered viable, both in terms of cost and in producing results equivalent to those of commercial weather station sensors.
|
